Fisher information matrix-based nonlinear system conversion for state estimation

In practical target tracking, a number of improved measurement conversion techniques have been developed and proofed to be superior to the standard (extended) Kalman filtering (KF) in Cartesian coordinates. The framework of conversion technique exhibits fundamental pros and cons and therefore associated with different performance as pointed out in. In this paper, we show that, based on the Fisher information matrix (FIM) which can be evaluated approximately using state estimates online, instead of the usual measurement conversion, an equivalent linear dynamics can be reconstructed from a general nonlinear form, thus even the standard KF can be applied theoretically. The proposed approach is explicitly free of the fundamental limitations of traditional measurement conversion. Simulation results are provided by comparison with a state-of-art conversion method with the so-called optimal linear unbiased estimate presented in.